Nonoccupational HIV post-exposure prophylaxis: a 10-year retrospective analysis

Abstract

Background

We conducted a retrospective analysis of administration of nonoccupational HIV post-exposure prophylaxis (nPEP) in a single centre where tracing and testing of the source of exposure were carried out systematically over a 10-year period.

Methods

Files of all nPEP requests between 1998 and 2007 were reviewed. Characteristics of the exposed and source patients, the type of exposure, and clinical and serological outcomes were analysed.

Results

nPEP requests increased by 850% over 10 years. Among 910 events, 58% were heterosexual exposures, 15% homosexual exposures, 6% sexual assaults and 20% nonsexual exposures. In 208 events (23%), the source was reported to be HIV positive. In the remaining cases, active source tracing enabled 298 HIV tests to be performed (42%) and identified 11 HIV infections (3.7%). nPEP was able to be avoided or interrupted in 31% of 910 events when the source tested negative. Of 710 patients who started nPEP, 396 (56%) reported side effects, among whom 39 (5%) had to interrupt treatment. There were two HIV seroconversions, and neither was attributed to nPEP failure.

Conclusions

nPEP requests increased over time. HIV testing of the source person avoided nPEP in 31% of events and was therefore paramount in the management of potential HIV exposures. Furthermore, it allowed active screening of populations potentially at risk for undiagnosed HIV infection, as shown by the increased HIV prevalence in these groups (3.7%) compared with a prevalence of 0.3% in Switzerland as a whole.

Introduction

The protective effect of nonoccupational HIV post-exposure prophylaxis (nPEP) against HIV transmission has been demonstrated in animal studies [1,2], trials on the prevention of vertical transmission from mother to newborn [3,4] and case–control reports after needlestick injures in healthcare workers [5,6]. Although Centers for Disease Control and Prevention (CDC) guidelines on nPEP were issued in 2005 [7], many countries around the world have been prescribing it for more than a decade [8–13]. In Switzerland, national recommendations have existed since 1997 [14,15]. In most centres, infectious diseases specialists or emergency physicians are responsible for nPEP, although any primary care physician can prescribe this treatment. The large nPEP cohort studies published to date predominantly involved populations of men having sex with men (MSM) [16–20] and victims of sexual assaults [11–13,21,22], who may not always be representative of populations seen in other centres around the world with different sociodemographics. We conducted a retrospective analysis on nPEP requests and management since its implementation in our centre 10 years ago.

Subjects and methods

Study design

We reviewed the files of all nPEP requests after potential HIV exposure between January 1998 and December 2007 in our HIV out-patient clinic at the University Hospital of Lausanne, Switzerland. The clinic acts as a primary care hospital for the local population of 650 000 persons. At night or during the weekend, exposed patients are seen in the emergency department and then referred to our clinic for follow-up. All subjects were eligible if exposure occurred outside the healthcare environment and met indications for nPEP prescription. Data were collected prospectively throughout the study period according to an institutional standardized procedure. Approval was obtained from the local ethics committee. Informed consent was not required.

Subjects

At-risk exposure was defined as unprotected receptive or insertive anal or vaginal intercourse, receptive oral sex with ejaculation, equipment sharing among injecting drug users (IDUs) and other situations where infectious body fluids came into contact with a mucous membrane or non-intact skin. The following situations were not considered to pose a risk of HIV transmission: protected sex, receptive oral sex without ejaculation, human bites without contact with the assaulter's blood, exposure of intact skin to body fluids or needlestick injuries in public settings unless there was a suspicion that the needle had been used recently (<1 h prior to exposure).

HIV status assessment of source persons

When the source was reported to be HIV infected, an attempt was made to confirm their HIV status by contacting the treating physician and, if contact was established, to collect information about the latest viral load as well as any ongoing antiretroviral treatment (ART). We did not perform HIV testing to confirm the serological status of reported HIV-positive source persons. When the HIV status of the source was unknown, index patients were strongly encouraged to contact and ask the source person to present at our centre for free anonymous HIV testing.

nPEP indications

Antiretroviral prophylaxis was considered for all patients exposed to a reported HIV-infected source within 72 h after exposure. After an update to national guidelines in 2006, nPEP was no longer prescribed when the source of exposure had an HIV viral load <50 copies/mL while taking ART for more than 6 months [15]. When the HIV status of the source could not be determined, nPEP was offered if the source subject belonged to a group at risk for HIV infection [a sexual assaulter, a man having sex with men, an IDU or a person from a high (>1%) HIV prevalence country]. Commercial sex workers, although not specifically mentioned in our national guidelines, were considered at risk for HIV infection when identified by the client as an IDU or coming from a high-prevalence country.

Drug regimen

For most participants, the drug regimen consisted of either zidovudine (ZDV) 300 mg and lamivudine (3TC) 150 mg twice daily plus nelfinavir (NFV) 1250 mg twice daily (from 1998 to 2007); or the same doses of ZDV and 3TC plus a fixed dose of lopinavir (LPV) 400 mg and ritonavir (RTV) 100 mg twice daily (from 2007 onwards). When an HIV-positive source was already on ART, the drug regimen was adapted to the history of treatment, ongoing therapy and available drug resistance analysis. The regimen was also modified to avoid potential drug interactions with concomitant medications. Prophylaxis was given for 28 days but was stopped earlier if the source subject tested HIV negative or the exposed patient was found to be positive at baseline testing.

Follow-up

At the first visit, demographic data were collected from exposed patients as well as information on the nature of exposure and risk factors for HIV infection for themselves and for source subjects. When nPEP was prescribed, a second visit was planned 2 weeks later to ascertain drug adherence and tolerance. Risk-reduction counselling was provided on each visit. Complete blood count and renal and liver function tests were assessed at baseline and at week 2. For all participants, antibody and p24 antigen HIV testing was offered at baseline and was repeated at 3 and 6 months. From 1998 to 2006, a third-generation assay (Roche Cobas Core anti-HIV 1+2+O EIA; Roche Diagnostics GmbH, Mannheim, Germany) combined with a p24 antigen assay (Roche Cobas HIV Ag) was performed, whereas from 2006 onwards, a fourth-generation assay (Cobas HIV Combi^®; Roche) was used. From 2006 onwards, the 6-month test was no longer performed following an update of our national guidelines [15]. When the source of exposure was found to be HIV negative, the decision to conduct follow-up HIV testing was left to the physician's discretion when there was a suspicion that the source might be in the preseroconversion window period.

Statistical analysis

A descriptive analysis of demographic data, the nature of exposure and risk factors for HIV infection was performed. Exposed subjects were categorized into risk groups. The likelihood of being able to contact and test the source of exposure was determined in each risk category of exposed patients by univariate analysis. We used Student's t-test when continuous variables were normally distributed and the Mann–Whitney U-test for skewed distributions. Categorical variables were analysed using Fisher's exact test. Data were analysed using stata 10.0 (Stata Corporation, College Station, TX, USA).

Results

Study population

Between 1998 and 2007, 1233 consultations for potential HIV exposure were recorded. A marked and steady increase was noted in the number of consultations per year, rising from 20 in 1998 to 196 in 2007 (+850%). Of these, 27 occurred in the healthcare setting and were therefore excluded. One hundred and thirty-eight consultations were also excluded from analysis because of missing data (90 cases), absence of exposure (34) and refusal of medical care by the subject (14). Among the remaining 1068, 158 exposures did not meet indications for nPEP prescription (Fig. 1). Overall, 910 events involving a total of 867 persons were included in the final analysis. Eight hundred and thirty individuals requested nPEP once, 31 twice and six three times (two MSM, one IDU, one commercial sex worker, one client of a commercial sex worker and one heterosexual not belonging to any risk group). Subjects were predominantly male (64%) and from countries of low (<0.5%) HIV prevalence (84%). The median age was 30 years (range 14–87 years). Fifty per cent of subjects did not belong to any known risk groups for HIV infection. The other 50% consisted of clients of commercial sex workers (16%), MSM (15%), IDUs (6%) and commercial sex workers (3%). Housekeepers, who frequently sought care after injuries from needles left in trash bags, and police officers, who were exposed to infectious body fluids during violent arrests, accounted for 2 and 3% of the subjects, respectively. Four per cent were stable partners of HIV-infected persons. Excluded subjects differed from those included in the analysis in the following ways: they were more likely to be older than 40 years, more likely to be exposed through nonsexual routes, and less likely to be IDUs and clients of commercial sex workers but more likely to be exposed as housekeepers.

Figure 1.  Eligibility and outcomes. nPEP, nonoccupational post-exposure prophylaxis.

Type of exposure

Of 734 sexual exposures, 527 (72%) involved heterosexual contact and 132 (18%) homosexual contact (see Table 1). Proportions of anonymous sexual contacts were similar in heterosexual and homosexual subjects (62 and 61%, respectively). Fifty-eight sexual assaults were also registered.

Table 1.   Type of exposure

Variable	nPEP (n=710)	No nPEP (n=200)
Data shown are n (%). ^* Three events involved mixed heterosexual contact and IDU equipment sharing. ^† ^† Involving 52 female and six male subjects. ^‡ ^‡ Defined as suspected unwilling sexual contact after alcohol or drug abuse. ^§ ^§ In cases of mixed exposures, only the highest risk was considered. ^¶ ^¶ Infectious body fluids on broken skin (42 cases) or on a mucous membrane (eight). IDU, injecting drug user; nPEP, nonoccupational post-exposure prophylaxis.
Route of exposure
Sexual	563 (79)^*	171 (85)
Nonsexual	147 (21)	29 (15)
Sexual exposure
Heterosexual	385 (54)	142 (71)
Missing data	67 (9)	36 (18)
Anonymous/casual contact	258 (36)	72 (36)
Stable partner	60 (8)	34 (17)
Homosexual	113 (16)	19 (9)
Missing data	24 (3)	9 (4)
Anonymous/casual contact	76 (11)	4 (2)
Stable partner	13 (18)	6 (3)
Sexual assaults^†	52 (7)	6 (3)
Doubtful cases^‡	13 (2)	4 (2)
Types of sexual intercourse^§
Vaginal insertive	218 (31)	78 (39)
Vaginal receptive	181 (25)	58 (29)
Heterosexual anal insertive	6 (<1)	1 (<1)
Heterosexual anal receptive	11 (1)	0
Homosexual anal insertive	53 (7)	7 (3)
Homosexual anal receptive	35 (5)	7 (3)
Homosexual anal insertive and receptive	6 (<1)	1 (<1)
Homosexual anal unspecified	13 (2)	5 (2)
Oral receptive with ejaculation	15 (2)	5 (2)
Low risk (none of the above)	12 (2)	5 (2)
Doubtful intercourse^‡	13 (2)	4 (2)
Condom use
Missing data	14 (2)	7 (3)
No condom	250 (35)	69 (34)
Condom breakage	266 (37)	87 (43)
Condom slippage	33 (5)	8 (4)
Nonsexual exposure
Percutaneous injuries	61 (8)	7 (3)
IDU equipment sharing	46^a (6)	1 (<1)
Human bites	7 (1)	7 (3)
Others^¶	36 (5)	14 (7)

The majority of the 179 nonsexual events were related to needlestick injuries (37%) and IDU equipment sharing (25%).

Characteristics of the source of exposure

In 208 episodes (23% of 910 eligible requests), the source was reported to be HIV positive, and in 187 episodes the HIV-positive status could be confirmed. Among those for whom information was available, more than half were not under ART and had a detectable viral load at the time of exposure. In 702 events (77%), the HIV status of the source subjects was unknown. In these cases, 298 (42%) source persons could be tested and 11 new HIV infections were diagnosed (see Table 2).

Table 2.   HIV characteristics of the source of exposure at baseline and after HIV status assessment

Characteristic	n (%) (n=910)
^* Including 292 persons and six needles. ^† ^† Including seven persons (two commercial sex workers, three men who have sex with men and two persons from countries of high HIV prevalence) and four needles.
Reportedly HIV positive at baseline	208 (23)
HIV status confirmed	187 (20)
HIV status not confirmed	21 (2)
Ongoing antiviral treatment
Missing data	71 (8)
Yes	52 (6)
No	85 (9)
Viral load
Missing data	114 (12)
Undetectable	29 (3)
< 1000 copies/mL	10 (<1)
1000–100 000 copies/mL	42 (5)
>100 000 copies/mL	13 (1)
Unknown HIV status at baseline	702 (77)
Tested	298 (33)^*
HIV negative	287 (31)
HIV positive	11 (1)^†
Not tested	404 (44)
Reason unknown	15 (2)
Source untraceable	272 (30)
Source refused	58 (6)
Exposed patient refused	56 (6)
Risk considered too low	3 (<1)

The likelihood of being able to contact and test the source varied significantly across risk categories. Police officers were more likely to have their source found and tested compared with non-police officer subjects (57 vs. 32%; P<0.001). Conversely, IDUs, MSM and housekeepers were less likely to have their source tested than non-IDUs (4 vs. 34%; P<0.001), non-MSM (24 vs. 34%; P=0.02) and nonhousekeepers (10 vs. 33%; P=0.02), respectively. No difference was seen for commercial sex workers (27% of sources tested) and clients of commercial sex workers (32%). Heterosexual subjects had their contacts tested more often than did MSM (38 vs. 24%; P=0.001).

PEP

The median time to consultation was 17 h after the exposure. Five hundred and forty-seven participants (60%) sought care within 24 h and 747 (82%) within 48 h. Among 910 eligible events for nPEP, it was received in 710 cases (78%) (Fig. 1). Twenty-six persons received nPEP twice during the study time, while five patients had three nPEP courses. Of note, of 258 cases of nPEP given for heterosexual contact with an anonymous or a casual source, 151 (58%) were prescribed although the source did not belong to any risk group for HIV infection.

Among 710 patients who initiated therapy, 423 (60%) completed nPEP and 117 (16%) were lost to follow-up. Among the remaining 170, prophylaxis was mainly interrupted because the source tested HIV negative (108 cases) or the treatment was not tolerated (39).

Overall, testing of the source person and obtaining a negative result avoided the initiation or completion of unnecessary nPEP in 283 requests (31%). In four cases, the patient decided to continue nPEP despite the source's negative result. The rate of avoided nPEP varied across types of exposure to HIV and was significantly correlated to the ability to find the source person (P<0.001) (Fig. 2).

Figure 2.  Rates of initial post-exposure prophylaxis (PEP) prescription and PEP avoidance/interruption when the source was tested HIV negative, according to type of exposure.

Drug regimens

Out of 710 nPEP prescriptions, ZDV+3TC+NFV was used in 548 cases (77%) and ZDV+3TC+LPV/RTV in 108 (15%). Forty-one subjects received various combinations of other antiretroviral drugs, and for 13 details of the nPEP regimen were not available. Of 620 participants for whom data were available, 396 (64%) reported side effects, mainly gastrointestinal disturbance (325 cases) and fatigue (189). At the week 2 visit, new-onset laboratory abnormalities, including leucopenia, thrombocytopenia, acute renal failure, hepatitis and pancreatitis, were seen in 41 subjects. They were all grade 1 or 2 toxicity except for four cases of grade 3 and 4 liver toxicity with the ZDV/3TC/NFV combination. One of these was attributed to hepatitis C virus seroconversion. Liver tests spontaneously improved after nPEP interruption, without hospitalization. Overall, 18 participants changed drug regimen and 39 stopped nPEP because of drug toxicity. The only differences between the two regimens were a higher frequency of headaches (P=0.02) and gastrointestinal disturbance, which did not reach statistical significance, in the ZDV/3TC/NFV group (Table 3).

Table 3.   Adverse events according to main drug regimens

Adverse event	Drug regimen
	ZDV+3TC+NFV (n=548)	ZDV+3TC+LPV/r (n=108)	P-value
Data are n (%). All adverse events are grade 1 or 2, unless otherwise indicated. ^* One case was attributed to hepatitis C virus seroconversion. 3TC, lamivudine; LPV/r, lopinavir/ritonavir; NFV, nelfinavir; ZDV, zidovudine; NS, not significant.
Missing data	71 (13)	13 (12)
Overall adverse events	316 (58)	49 (45)	0.01
Gastrointestinal intolerance	261 (48)	42 (39)	0.07
Fatigue	140 (26)	33 (30)	NS
Headaches	66 (12)	5 (5)	0.02
Vertigo	4 (<1)	0	NS
Overall laboratory abnormalities	35 (6)	3 (3)	NS
Leucopenia	16 (3)	0	NS
Thrombocytopenia	2 (<1)	0	NS
Acute renal failure	6 (1)	3 (3)	NS
Hepatotoxicity	12 (2)	0	NS
Grade 3	1	0
Grade 4	3^*	0

HIV tests

Among 910 eligible events, 865 (95%) exposed persons were tested at baseline, 468 (51%) had a second test at 3 months and 202 (22%) had a third test at 6 months. Among 287 subjects exposed to an HIV-negative source, 61 (21%) came back for a second test vs. 147 of 219 subjects (67%) exposed to an HIV-positive source and 260 of 404 subjects (64%) exposed to a source of unknown HIV status.

At baseline, two exposed subjects were HIV positive (0.2%). Upon follow-up, two HIV seroconversions were observed, neither of which was attributable to nPEP failure. The first case involved a 24-year-old homosexual man whose condom broke during anal insertive intercourse with a man who tested negative at that time. No nPEP was prescribed. HIV seroconversion was diagnosed 2 months later when he presented with acute retroviral syndrome, 3 weeks after unprotected anal receptive sex with an anonymous partner. The second case was a 24-year-old female IDU who was exposed through vaginal contact with an HIV-infected source. PEP was prescribed and completed. An HIV test was negative 4 months after exposure but seroconversion occurred 3 months later after ongoing IDU equipment sharing.

Discussion

This study is among the largest cohort studies on HIV nPEP. It includes a population of subjects potentially exposed to HIV through various routes, both sexual and nonsexual. Our results demonstrate the feasibility and efficiency of a strategy based on active tracing of the source of exposure as a means to reduce unnecessary antiretroviral prophylaxis.

Current CDC guidelines recommend the prescription of nPEP in cases of exposure to a known HIV-infected source [7]. A study by Pinkerton et al. [23] concluded that nPEP was only cost-effective in cases where men reported receptive anal intercourse with an infected partner. However, HIV transmission by partners of unknown HIV status has already been reported in this context [24]. In cases of nonoccupational exposures, especially for anonymous sexual contacts, the HIV status of the source is often unknown, as was the case in our study for 77% of events. CDC guidelines do not recommend for or against the use of nPEP in these situations but favour a case-by-case approach in which risks and benefits are weighed [7]. Swiss national guidelines recommend prophylaxis in situations where the source person belongs to a high-risk group for HIV infection (MSM, IDU, individuals from high HIV prevalence areas and sexual assaulters) [15]. For this reason, in most nPEP studies published to date, antiretroviral prophylaxis has been provided for both documented and high-risk potential exposures to HIV [12,13,16–20]. The only way to overcome this problem and avoid unnecessary prescription of antiviral prophylaxis is to test the source subject whenever possible, as stressed by some guidelines [7,25]. Tracing and testing the source person has already proved feasible and cost-saving [20,26]. In a previous report based on a smaller sample of the same cohort, this strategy was found to reduce the number of nPEP prescriptions by 28% [26].

In our study, source persons of unknown HIV status could be tested in 42% of events, a proportion significantly higher than previously reported (7–16%) [16,20]. The reason why we obtained such a high rate of source persons presenting for testing was probably related to the proactive way in which we explained to the exposed patients the benefits of avoiding or interrupting nPEP if the source was tested negative for HIV. These included not having to be exposed to antiretroviral drugs with known side effects for 28 days and the financial benefit of not paying for the entire course of nPEP (in Switzerland, the cost of nPEP is charged directly to the patient and then partially reimbursed through medical insurance). This approach allowed us to avoid or interrupt unnecessary nPEP in 31% of eligible events, contributing to reduced healthcare costs, potential drug toxicity and anxiety for the exposed person. However, this strategy was more efficient for some types of exposure (heterosexual contacts and sexual assaults) than for others (homosexual contacts and IDU equipment sharing), underlining the difficulty of reaching the source of exposure in some settings. Surprisingly, commercial sex workers and clients of commercial sex workers were not less likely to have their source tested than the rest of the study population. The difference between heterosexual and homosexual subjects could not be explained by differences in frequency of anonymous contacts, as one might have expected. However, it is possible that the definition of anonymous contacts did not encompass the same realities in the two groups, as many anonymous MSM contacts occurred in bathhouses with truly untraceable contacts.

Testing the source also allowed us to detect 11 undiagnosed HIV infections. The HIV prevalence of the source population of unknown HIV status was therefore 3.7%, a proportion 10 times higher than that reported in the general population in Switzerland [27]. When source subjects that were reported to be HIV positive were included, the prevalence increased to 24%, which is consistent with other reports [13,17]. Sixty-two per cent of those for whom information was available were not treated and 69% had a detectable viral load. These data underscore the risk of undiagnosed and untreated HIV infection in the population of source subjects and therefore support the prescription of nPEP in cases of exposure to persons of unknown HIV status belonging to high-risk groups. However, in this study, a significant proportion (58%) of subjects reporting heterosexual contact with an anonymous or a casual partner were prescribed nPEP, although the source was not reported to belong to any risk group for HIV infection. Although this practice is not endorsed by our national guidelines, antiretroviral prophylaxis was provided in these cases because the source was reported to have multiple sexual partners and believed to be at risk for HIV infection.

We observed two seroconversions. Neither was linked to nPEP failure, as infection occurred after ongoing risk behaviour. The fact that one of the two patients was not offered prophylaxis at the time of consultation does not call into question our policy to withhold nPEP when the source is tested negative. Indeed, fourth-generation tests have recently been shown in percutaneous occupational exposures to detect p24 antigen during acute HIV infection when antibodies are still undetectable [28]. The absence of nPEP failure, however, cannot be considered proof of its efficacy as the sample size was too small to allow assessment of such a rare phenomenon.

A major limitation of our study was the high drop-out rate throughout the follow-up period. Overall, 16% of patients for whom nPEP was initiated never came back for assessment of regimen completion and drug toxicity and 49% of all participants never had a second HIV test at 3 months. This problem is encountered in most studies on nPEP given the particular context of confidentiality issues surrounding anonymous sexual contacts and psychological sequelae following sexual assaults, all of which may explain why some patients were lost to follow-up. Indeed, our drop-out rate was consistent with those reported elsewhere (25–74%) [12,13,17,18,21,22]. Of note, many drop-outs involved subjects exposed to an HIV-negative source, a situation in which follow-up testing is not mandatory. Another limitation was the retrospective aspect of our analysis and the fact that data were limited to those that could be obtained from case note reviews. However, files were often complete and only a minority of nPEP requests could not be analysed because of missing data (7%).

PEP prescription in cases of exposure to a source of unknown HIV status is an everyday challenge for most reference centres world-wide. Although available HIV prevalence data for high-risk groups favour the use of prophylaxis in these situations, testing the source person probably represents the best and most cost-effective way to avoid unnecessary exposure to antiviral prophylaxis. It also represents a unique opportunity to screen a difficult-to-reach population engaging in practices carrying a high risk for HIV infection. When the HIV status of the source cannot be determined, the decision to offer prophylaxis should be based on an individual evaluation of risk factors given the high prevalence of undiagnosed HIV infection in this population.

Acknowledgments

We thank Serge Gallant, Sophie Farine, Véronique Fardel, Véronique Nicklas and Vreneli Waelti for their indispensable help in collecting clinical data throughout the study period.

Author contributions: F.T. had full access to all data and takes responsibility for the accuracy of the data analysis. M.C. was responsible for the concept and design of the study. F.T. analysed the data and drafted the manuscript. M.C., V.E. and T.D. were involved in critical revision of the manuscript. V.E. provided statistical expertise.

Financial support: None.

Potential conflicts of interest: F.T. has received travel grants from Tibotec/Janssen-Cilag AG. T.D. has received travel grants from Merck Sharp & Dohme and Tibotec/Janssen-Cilag AG. M.C. has received travel grants from Abbott, Boehringer-Ingelheim, Gilead and Roche. V.E. has no conflict of interest.